Modeling Subcommittee Meeting
Chesapeake Bay Program Office; Annapolis, MD
December 4, 2007

*A11 presentations can be accessed at:

http://www.chesapeakebav.net/calendar.cfm?EventDetails=8031&DefaultView=2.

I.	Announcements and Amendments to the Agenda	Linker

•	Lewis Linker began the meeting at 10:00am. Announcements were made and the
meeting's agenda was reviewed.

•	The Phase 5 Model will receive a final set of inputs by February 1, 2008. We'll as
the Subcommittee for final review of Phase 5 after we receive these final inputs.

•	There may be some difficulties with the bed model in the water quality model. This is
an issue that the subcommittee may need to discuss at their January meeting.

II.	Overview of WQSTM and Living Resource Model Linkage	Cerco

•	Carl Cerco, USACE, discussed the progress that has been made in the development of
the filter feeder model simulations for oysters and menhaden and he presented an
approach to linking the Water Quality Sediment Transport Model (WQSTM) with the
trophic models.

Oysters

•	They have the 2004 starting population and the distribution from the demographic
model, as well as useful information on mortality and catch.

•	There are, however, some issues. These include the following:

o The demographic model uses multiple size classes (market, small, and spat),
o The biomass and distribution are substantially different from previous
estimates.

•	The following decisions need to be made:

o What biomass estimate should we start with?
o How should we distribute it?

o How should we deal with market-size vs. smalls? (Combine into one market-
size class? Have two oyster size classes?)

•	Q: Why does the size class matter?

o A: If you take a filtration rate for unit biomass, it will be higher for smaller
oysters.

•	Q: Is there an estimate of mortality from disease and mortality from harvesting at the
Chesapeake Bay segment level?

o A: There are natural mortality estimates, including disease, by three salinity
classes that could be broken out into Chesapeake Bay segments. There aren't
currently any estimates for mortality from harvesting on a CB segment level
that the community feels confident about. There is information on landings by
area, but these are not bar-specific estimates of the population size before the

1


-------
harvest. Well need to use an extremely aggregated scale for estimates of
harvest mortality

•	Summary:

o The model is in place.

o They've done it before.

o It should be straightforward.

Menhaden

•	This is a work in progress. A lot of work still remains on the bioenergetics aspects.

•	Menhaden are represented in the simulation as discrete schools with specified
movement through the Bay. Schools have specific properties including number of
individuals and age structure. The influence of the school on water quality will be
determined by bioenergetics models and VIMS results.

•	Carl showed the group a visualization of three schools of menhaden swimming
around the Chesapeake Bay. This visualization was on the 12,000 cell grid and is still
a work in progress.

•	There's a lot of uncertainty regarding ecosystem response to changes in menhaden
population and the ability of ICM to reproduce this response.

o Will increased menhaden grazing mean more predation or a switch in
predation?

o Does the zooplankton population go down if the menhaden population goes
up?

•	Q: There already is a published bioenergetics model for menhaden by Durbin and
Durbin. Is this being looked at?

o A: Yes. They do not want to develop a new bioenergetics model. Instead, they
are trying to use what is already out there.

•	Q: Menhaden tend to be concentrated in areas with higher chlorophyll. Is this being
incorporated into the model?

o A: No. In this model, menhaden don't chase chlorophyll. In the long run, they
would like that to happen. However, the technology is not there yet in this
model.

•	Q: Is there a sense of the number of menhaden schools that would ultimately be
simulated?

o A: There are not that many schools, so there will not likely be many more
added to the simulation.

•	There are two components of the menhaden population that should be modeled:
schools and residents. However, it is hard to tell what proportion of the population is
made up of residents.

•	We have about three menhaden schools we'll simulate. Do we want a shallow water
school that we can turn on and off to see what effect the juveniles in shallow water
have on water quality?

•	Summary:

o A lot of development remains.

o Not at all straightforward.

Coupling Ecological Models with Eutrophication Models

2


-------
•	The proposed approach presented by Carl is to:

o Couple the Chesapeake Bay CE-QUAL-ICM eutrophication model with the
existing Ecopath with EcoSim (EwE) network model for the Chesapeake Bay
upper trophic levels,
o Use the Ecopath model presented by Jim Hagy in his PhD dissertation,
o The key information transmission would be ICM primary production and
detritus concentrations.

•	Ecopath is steady state, like a spread sheet. For a first order look, this is likely a good
approach.

•	Questions that need to be addressed include:

o How should we process spatially and temporally detailed ICM information for

use with Ecopath?
o What are the commonalities between ICM and EwE?
o How should the two models be interfaced?

•	Ecopath will have to be rebalanced based on new values from ICM.

•	The production and detritus values assumed by Hagy in his base run are not
necessarily consistent with our ICM existing conditions, thus initial adjustments may
be required before the scenarios are run.

•	For higher trophic levels, the process is in place and proofed, but it still needs to be
transferred to the 50,000 cell grid and further refined. Some effort will be required.

•	The issue of higher trophic level simulation can be set aside until post-April.

•	If there are better biomass estimates from Ecopath or EcoSim, they should be sent to
Carl.

III. Oyster Demographics Model	Weber

•	Ed Weber, Versar Inc., presented information on an oyster demographics model that
was developed in support of a programmatic EIS for oyster restoration in order to
compare alternative restoration plans.

•	This model is not necessarily good for total oyster biomass estimates. Instead it
focuses on biomass changes due to management alternatives.

•	Estimates are best on a large scale. The model is spatially explicit to the level of
8,480 polygons based on hard bottom surveys, which are then aggregated up to higher
levels. They are less comfortable with estimates that are more specific than those for
individual states, such as for individual bay segments.

•	The starting populations used came from MD DNR and VIMS/VMRC surveys.

•	For the key physical conditions that were simulated, USGS historical data was used
for annual precipitation and Bay program monitoring from 1995-1999 was used for
salinity.

•	Recruitment data was based on the ratio of spat to adults in the MD DNR fall survey.
It included effects of salinity and episodic large spat sets. Recruitment was then
redistributed according to the larval-transport model.

•	Growth data came from four different sources, including Coakley (ages 0-2), Paynter
(adjusted to nearest year), and Rothschild et al. (pre disease) for MD and Roger Mann
for Virginia.

3


-------
•	Natural mortality is based on the MD fall survey (1991-2006) and uses ratios of
"recent" boxes to live oysters.

•	1,000 simulations were run for each alternative. The alternatives that were analyzed
were:

o Alt. 1- Status quo (225 million seed, 215 acres planted per year)
o Alt. 2- Enhanced restoration (2 billion seed, 9,012 acres planted annually),

there was an alt. 2a and an alt. 2b that differed in where the seed was placed
o Alt. 3- Harvest moratorium

•	The model looked at the results in year 10. According to the model, if you ramp up
restoration, you get a higher population. If there is instead a harvest moratorium,
population increases somewhat, but not to the extent that it did with restoration.

There was a sense that a significantly longer simulation period (beyond 10 years)
would enhance the outcome of the moratorium scenario as a key problem is that we
start with low current oyster biomass and more time would allow the moratorium
approach to build up more biomass over time.

•	The greater the population, the more harvest makes a difference. Low population
mortality is dominated by disease.

•	Challenges and data limitations:

o Surveys not designed to estimate abundance

-	Error in starting populations is unknown

-	Estimation of recruitment is non-standard
Cannot use historical data to validate the model

o Insufficient data to model changes in habitat (also shell budgets)
o Limited spatial data, especially for Virginia
o Harvest rates are unquantified

•	Conclusions

o High natural mortality dominates

o Harvest effects biomass increase as the stock recovers

o Ongoing enhanced restoration may help, but the EIS goal probably cannot be

met in ten years
o Input data sets create unquantifiable uncertainty

o Asian oysters will not be modeled, but model sensitivity runs will be used to
inform decisions

o Also a useful tool to simulate different harvest and sanctuary strategies,
effects of drought, etc.

•	Q: Could this model be used for purposes beyond this EIS?

o A: Yes, it can go beyond the EIS. It could be useful for what-if scenarios.

•	This report is not yet finalized; however, Ed offered to send the draft report to anyone
who is interested. Once the report is finalized, it will likely be posted on the internet.

•	There is still a question as to which biomass estimate CBP should use.

IV. Oyster Restoration Optimization (ORO) Model	North

•	Elizabeth North, UMCES, provided an overview of the Oyster Restoration
Optimization model (ORO). The objective of this project was to create a flexible
ecosystem-based decision-making tool to support oyster restoration and management.

•	This model is in the proof-of-concept stage.

4


-------
•	The ORO model is designed to:

o Track the growth and mortality of hatchery-produced oysters planted at
different sites.

o Estimate benefits (e.g., oysters impact on water quality, production of future

spawners, and harvest potential),
o Determine optimum locations to place oysters that maximize desired benefits
given current constraints.

•	Approach:

o Predict growth and survival of hatchery produced oysters with a demographic
model

o Calculate ecosystem services using a 3D circulation and water quality model
o Estimate spawning success using fecundity and larval transport models
o Incorporate costs and constraints of restoration

•	The ORO model is a spatially explicit model that uses the water quality model grid
structure. It is designed to identify optimum locations to place hatchery-produced
oysters in order to maximize desired benefits.

•	The Chester River, Choptank River, and Tangier Sound were used as test sites. The
model tracks the size, abundance, and benefits of oysters in each area.

•	The model must be run multiple times for each region (in this case, a region is a cell).
In this example, there were 15 regions with 5 oyster abundances in each region, so the
model had to run 75 simulations.

•	The RISKOptimizer solution package was used, which is an Excel Add-in. This
package incorporates environmental and biological variability, as well as constraints
(costs of restoration and available funds, hatchery production limitations, habitat
availability).

•	Application as a decision-support tool:

o Ecosystem-based: Estimates interactions between a single species and the

ecosystem and incorporates uncertainty,
o Tactical: The ORO model (if enhanced) could provide quantitative
information about the trade-offs associated with different management
decisions.

o Strategic: Same framework can be used to understand how spatial changes in
harvest would affect the ecosystem and oyster populations (i.e., negative
impacts of harvest could be minimized).

•	Potential next steps: incorporate predictions from higher-resolution water quality
model; enhance optimization routine; improve parameterization of larval and adult
mortality; validation; improve costs/constraints; and refine harvest mortality.

•	It would be useful to have error bars on predictions that incorporate several years of
environmental variability- right now it shows just one year.

•	If this model were to be expanded, one suggestion was to have it just look at regions
and not cells in order to decrease the number of model runs needed.

•	Funding for this project is over.

V. Examining the Effect of Menhaden on Water Quality Brush and Lynch

5


-------
•	Mark Brush and Patrick Lynch, VIMS, presented their progress on the last year of a
three year research project that looked at modeling Atlantic menhaden in support of
nutrient and multispecies management in the Chesapeake Bay.

•	As part of this project, there were a set of feeding experiments. Experiment 1 focused
on diet selectivity, functional response, and excretion rate (DIN, DON) and
experiment 2 looked at density dependence.

•	The experiments were conducted in 2006 and 2007 (five pilot experiments were
conducted in 2005).

•	Patrick described the methods that were used in these experiments. See his
PowerPoint presentation for more details.

•	The objectives of these experiments were to analyze the following factors:

1.	Ingestion rates of phytoplankton (phytoplankton concentration)

2.	Excretion rates of nitrogen (phytoplankton concentration)

3.	Rates of net removal of nitrogen (via ingestion on phytoplankton).

•	Experiments were conducted separately on young of year (YOY) fish and age 1+ fish.
The only difference between these two sets of experiments was the number of fish per
tank.

•	In addition to the experiments designed to meet the above three objectives, an
additional experiment was conducted that included zooplankton in the treatment tank.
This experiment was identical in design to the other experiments, but with the
following exceptions: the duration was shortened to 3 hours; the YOY and age 1+
experiments were run simultaneously; zooplankton (predominately A. tonsa) were
placed in the treatment tanks; and the ammonia concentration differed (to, ti.s, t3).

•	The results of these experiments indicate that the age 1+ fish were willing to feed in
the experimental tanks, but that they just weren't feeding on phytoplankton. See
Patrick's PowerPoint presentation for more details on the results.

•	Q: In this experiment, YOY fed more on the phytoplankton community than the age
1+ fish. Why was there a difference?

o A: This may be due to differences in gill raker morphology and particle
retention. The phytoplankton culture that was used in this experiment
consisted primarily of very small phytoplankton, thus since the gap size
between the gill rakers was much larger in the age 1+ compared to the YOY,
the phytoplankton may have slipped right through the gill rakers in the age 1+
fish, but not in the YOY.

•	The implication of this for the Chesapeake Bay is that YOY may significantly impact
phytoplankton, while the age 1+ fish may have less of an impact. However, the age
1+ fish do have the potential to graze on mesozooplankton.

•	Based on this experiment, they can't really say whether there was a density dependent
effect on the filtration rate of the age 1+, but the experiment does indicate that as
density increases, YOY filtration rates tend to decrease.

•	Another part of this study involved looking at stable isotopes. For this part of the
experiment, they measured the isotopic composition of field-caught fish. The 2005
and 2006 samples have already been processed and analyzed. The 2007 samples have
not yet been completed. See Mark's PowerPoint presentation for some of the results.

•	Q: Are there significant variations in isotopic signatures for menhaden based on
location?

6


-------
o A: As you move down the Bay, you don't see much of a change in N. Carbon
changes slightly, but not too much. Particulate matter gets more positive as
you move towards the mouth of the Bay, but it is not really a big change.

•	Q: Are there significant variations in isotopic signatures for menhaden based on the
season?

o A: There was no discernable trend in the carbon, N, or particulate matter
signature based on the season.

•	Steps that are currently in progress for this study include:

o DON sample analysis
o Two manuscripts and data delivery
o 2007 isotope sample analysis
o Extraction and analysis of lipids from menhaden

•	This summer, there will be an experiment that will look at the time spent feeding.

•	Now that they have information on the functional responses and the density
dependence responses, they will begin updating the energetics model.

•	Mark Brush agreed to provide the subcommittee with another update in either January
or February 2008.

VI. Progress in the Ecopath/EcoSim Model	Townsend and Ma

•	Howard Townsend, NOAA CBO, and Hongguang Ma, NCBO-Versar, provided the
subcommittee with an update on the Chesapeake Bay Fisheries Ecosystem model
(CBFEM).

•	The ultimate goal is to use CBFEM (and other ecosystem management models) in a
process similar to single species stock assessment models. The CBFEM will be
developed, reviewed, applied, and updated on a regular cycle with oversight from a
technical committee.

•	Under the Information Quality Act (2000), NOAA has to have procedures in place for
ensuring and maximizing the "quality, objectivity, utility, and integrity" of
information (including statistical information) disseminated by federal agencies.

Thus, there is a need for technical review.

•	Progress towards technical review:

o Community input on development since 2002
o Living Resources Subcommittee Review (2005-2006)
o Ecosystem Modeling Technical Advisory Panel (EMTAP) data review (2006-
2007)

o Publication of CBFEM technical report in NOAA professional papers series

(planned submission in FY08)
o Publication of Chesapeake Bay Regional Estuarine Ecology Model

(CBREEM) as NOAA technical memorandum (planned submission FY08)
o Review of model by Center for Independent Experts (initiated in FY08)

•	NEMoW stands for National Ecosystem Modeling Workshop. This was a national
workshop that was held to standardize methodologies and approaches when using
ecosystem, bio-physical, and multispecies models. NEMoW products included: a
workshop report; an evaluation of models; recommendations for national EM
standards/guidelines for use and review; and recommendations for standardized
approaches.

7


-------
•	The Chesapeake Bay Fisheries Ecosystem Model was developed in cooperation
between NOAA CBO/Oxford, CRC, and UBC with support from many bay
researchers. It is a companion to the CB Fisheries Ecosystem Plan and it replicates
ecosystem history from 1950 to the present.

•	EwE Ecotracer is part of the Ecopath with EcoSim (EwE) approach and software
system, which includes Ecopath (for model parameterization based on mass-balance),
EcoSim (for time-dynamic simulation), and Ecospace (for time and spatial dynamics).
Ecotracer is integrated with the three major parts of EwE.

•	The purpose of the Chesapeake Bay Regional Estuarine Ecology Model (CBREEM)
is to generate historical patterns in primary productivity for EwE. CBREEM is a two-
layer simple hydrographic model that uses wind, rainfall, gage inflow, and relative
loading as inputs. The model interface shows spatial distribution of nutrients, salinity,
and chlorophyll a and has a user interface showing many parameters.

•	Results from CBREEM include Chla (used as nutrient loading forcing function for
EwE) and a comparison with biomass in EwE. The next step is to do more of a
statistical analysis.

VII. Adjourn

•	The meeting was adjourned at 3:00pm.

Subcommittee Members

Steve Bieber
Sally Bradley
Carl Cerco
Lee Curry
John Everett
Krystal Freeman
Kate Hopkins
Jacqueline Johnson
Bill Keeling
Lewis Linker
Hongguang Ma
Mike Naylor
Elizabeth North
Tom Parham
Howard Townsend
Jim Uphoff
Ed Weber
Jing Wu

and Participants

COG
CRC
USACE
MDE

Ocean Associates, Inc
CRC

UMCES/CBPO

ICPRB/CBPO

VADCR

EPA/CBPO

NOAA CBO

MDDNR

UMCES

MDDNR

NOAA CBO

MD DNR

Versar

UMCES/CBPO

sbieber@mwcog.org

sbradlev@chesapeakebav.net

cercoc@wes.army.mil

lcurrev@mde. state, md. us

iohneverett@oceanassoc.com

freeman.krvstal@epa.gov

khopkins@chesapeakebav.net

iiohnson@chesapeakebav.net

william.keeling@dcr.virginia.gov

linker.lewis@epa.gov

hongguang.ma@noaa.gov

mnavlor@dnr. state.md.us

enorth@hpl. umce s. edu

tparham@dnr. state.md.us

howard.townsend@noaa.gov

iuphoff@dnr. state, md.us

eweber@versar.com

iwu@chesapeakebav.net

On the phone:
Mark Brush
Patrick Lynch

VIMS
VIMS

brush@vims.edu

8


-------